Mesquite-associated soil and phyllosphere microbial communities differ across land-use types in drylands

Land-use change is profoundly altering the biodiversity and functioning of terrestrial ecosystems worldwide (Foster et al., 2003; Newbold et al., 2015). Drylands constitute 41% of the world’s terrestrial land surface and are among the most vulnerable terrestrial biomes to land-use change (Hoover et al., 2020). Limited water availability, low soil organic matter, and restricted plant growth in drylands can prolong ecosystem recovery after a disturbance (Steven et al., 2021) and can produce irreversible regime shifts in vegetation and soil (Bestelmeyer et al., 2015; Berdugo et al., 2022). For instance, overgrazing can lead to shrub encroachment and increased soil erosion (Van Auken, 2009; Condon et al., 2020), and urbanization can impact biodiversity via habitat reduction, simplification, and fragmentation (Groffman et al., 2014). The arid Southwestern United States is experiencing one of the fastest population growths in the country, leading to rapid and significant modification of land required for livestock grazing and urban areas (Baker et al., 2002; Stehfest et al., 2019).

Plant-associated microorganisms, both belowground and aboveground, are increasingly becoming the focus of land-use research due to their crucial roles in maintaining vegetation functions, health, and fitness (Barrios, 2007; Mueller et al., 2016; Laforest-Lapointe et al., 2017a). On the one hand, belowground soil microorganisms can reduce drought stress and promote plant nutrient acquisition (Rubin et al., 2017), as well as mediate plant community composition (Barberán et al., 2015; Buzzard et al., 2019) and biotic interactions between plants and herbivores (Biere and Bennett, 2013). In drylands, some soil microorganisms store and translocate nutrients between plants and biological soil crusts, leading to the increase of spatial and temporal access to scarce resources (Collins et al., 2008). However, some soil microorganisms can act as virulent plant pathogens and negatively impact plant health by causing a myriad of diseases such as root rot and powdery mildew (Fisher et al., 2012; Dietzel et al., 2019). On the other hand, the phyllosphere, defined as the aerial surface of plants, is home to up to 10²⁶ bacterial cells at a global scale (Lindow and Brandl, 2003) and is a reservoir of microorganisms in soil and atmosphere (Vacher et al., 2016). Phyllosphere microorganisms can influence plant growth and health by regulating hormone production (Bodenhausen et al., 2014) and by contributing to pathogen exclusion (Rastogi et al., 2013). Moreover, phyllosphere microorganisms have strong associations with plant functional traits (Kembel et al., 2014) and can promote plant diversity and terrestrial ecosystem productivity (Laforest-Lapointe et al., 2017b). Consequently, simultaneously considering both the soil and phyllosphere microbiomes is required to obtain a more complete picture of how plant microbiomes vary across land-use types.

The mesquite tree (Prosopis spp.) is a common shrub/small tree native to the Southwestern United States. Mesquite trees are managed for flour milled from pods (Nabhan et al., 2020), which contain polysaccharides that slow the digestion of sugar and increase insulin sensitivity (Brand et al., 1990). Mesquites have unique adaptations to arid ecosystems because their deep roots can absorb water far below the soil surface and their small leaves can decrease the amount of water loss. Mesquite trees exert a profound influence on neighboring vegetation and soil by providing vegetative cover and resources (Fredrickson et al., 2006). As a leguminous plant, mesquite trees have bacteria that fix atmospheric nitrogen (N), converting inorganic N forms available for plant uptake (Zahran, 1999). Therefore, studying the changes in mesquite-associated soil and phyllosphere microbiomes across land-use types would help us better understand the ecological consequences of land-use change on the growth and health of mesquite trees as well as develop management and restoration practices in drylands (Richard Teague et al., 2008; Bashan et al., 2012; Gornish et al., 2021).

In this study, we investigated the mesquite-associated soil and phyllosphere microbiomes (bacteria/archaea and fungi) in grazing versus natural areas and urban versus natural areas in the arid Southwestern United States. We hypothesized that (i) significant differences in the richness, community composition, and functioning of soil and phyllosphere microbiomes would be observed in grazing and urban areas compared to natural areas because soil properties, as well as the growth and health of mesquite trees, can diverge among land-use types and (ii) the patterns observed in (i) would be different between bacteria/archaea and fungi as well as between soil and phyllosphere because bacteria/archaea and fungi may adopt different ecological strategies relative to land-use type, and the impacts of land-use type can differ between soil and phyllosphere.

We collected soil and leaf samples from mesquite trees in 6 locations in the arid U.S. Southwest—in or near Tucson, AZ, USA—that correspond to different land-use types (natural, grazing, and urban; Figure S1). Catalina State Park (CSP) and Sabino Canyon (SC) are natural protected areas. CSP consists of deep, well-drained soils classified as Sasabe–Caralampi complex soils consisting of 45% Sasabe sandy loam and 30% Caralampi extremely gravelly sandy loam. The sampling area within SC is located adjacent to an ephemeral stream in the Coronado National Forest, with soil type similar to alluvium. Santa Rita1 (SR1) and Santa Rita2 (SR2) correspond to heavy (2,562 acres with a herd size of 436 grazing for a period of 8 days) and light (815 acres with a herd size of 77 grazing for a period of 64 days) grazing pastures within the Santa Rita Experimental Range. SR1 is comprised of coarse-loamy, mixed soils classified as Anthony series, a very deep well-drained soil formed by stratified alluvium, and SR2 has Lampshire–Chiricahua association soils, which is defined as a very cobbly loam. Reid Park (RP) and the University of Arizona (UA) campus are classified as urban land-use types with sparse, manicured, and nongrazed vegetation. RP is classified as Mohave series and urban land consisting of very deep, well-drained soils formed in mixed alluvium. UA soils are classified as Cave series consisting of very shallow, well-drained loamy soils formed in mixed alluvium. In summary, CSP and SC are natural locations, SR1 and SR2 are grazing locations, and RP and UA are urban locations.

Natural and grazing locations had similar native vegetation, including velvet mesquite (Prosopis velutina), blue palo verde (Parkinsonia florida), saguaro (Carnegiea gigantea), cholla (genus Opuntia), and creosote bush (Larrea tridentata). The grazing areas had more perennial grass cover (mainly genus Bouteloua) than the natural locations. Vegetation in urban locations included common species native to the Southwest but generally had more exotic species such as cork oak (Quercus suber), eucalyptus (Eucalyptus microtheca), mulberry (Morus alba), and walnut (genus Juglans).

In each location, 5 mesquite trees with pairwise distance of 10–20 m were sampled. For each mesquite tree, one soil sample beneath the canopy was collected within the root system at a depth of 10 cm, adjacent to the base of the tree. The leaves of mesquite trees are compound pinnate with 20–50 small oblong leaflets. For each mesquite tree, 10–15 leaflets (∼100 g) throughout the canopy (1.5 m from the soil surface) were randomly selected, and they were mixed into one composite sample. This resulted in a total of 30 samples (6 locations × 5 mesquite trees) for soil and leaf, respectively. All samples were collected from October 27 to October 29, 2019.

Both soil and leaf samples were individually placed in sterile Whirl-Pack plastic bags, stored on ice, and immediately transported to the laboratory. Soil samples were sieved through 2-mm mesh and homogenized. For each soil sample, a subsample was placed into a sterilized tube and stored at −80°C until molecular analyses were performed, and the remaining soils were air-dried overnight and used for chemical analyses. Phyllosphere microbial cells were collected immediately after collecting the leaf samples (see next).

To obtain an overview of soil chemical properties across locations, pH, electrical conductivity (EC), as well as total carbon (C) and nitrogen (N) contents were measured. Soil pH was measured with a ratio of 1:1 (w/v) soil to deionized H₂O using a FiveEasy Plus pH meter (Mettler Toledo, Columbus, OH, USA) in combination with an accumet Single-Junction pH electrode (Fisherbrand, Waltham, MA, USA). Soil EC was measured with a ratio of 1:5 (w/v) soil to deionized H₂O using a FiveEasy Plus pH meter combined with a Cond probe LE703 (Mettler Toledo, Columbus, OH, USA). To measure soil C and N concentrations, the samples were oven-dried at 105°C, manually ground with a mortar and pestle, and weighed at 1 g for combustion analysis using an Elementar Vario Max Cube (Elementar Americas, Ronkonkoma, NY, USA).

Soil genomic DNA was extracted using the Qiagen DNeasy PowerLyzer PowerSoil Kit (Qiagen, Hilden, Germany) according to the manufacturers’ instructions. Bacterial and archaeal communities were analyzed through PCR amplification of the V4 hypervariable region of the 16S rRNA gene using the 515-F (GTGCCAGCMGCCGCGGTAA) and 806-R (GGACTACHVGGGTWTCTAAT) primer pair (Caporaso et al., 2012). For fungal communities, PCR amplification of the first internal transcribed spacer (ITS1) region of the rRNA operon was performed using the ITS1-F (CTTGGTCATTTAGAGGAAGTAA) and ITS2 (GCTGCGTTCTTCATCGATGC) primer pair (Deshpande et al., 2016). The primers included Illumina adapters with the reverse primers having an error-correcting 12-bp barcode specific to each sample to allow for demultiplexing. PCR was conducted in 40 ml triplicate reactions per sample using 3 ml of extracted DNA, 3 ml of each primer, 20 ml of MyFi PCR Mix (Bioline, Taunton, MA, USA), and 11 ml of water. PCR consisted of an initial denaturing step at 95°C for 1 min, 35 cycles of amplification (95°C for 15 s, 60°C for 15 s, and 72°C for 15 s), and a final elongation step of 72°C for 3 min. Negative controls were included to detect potential contamination. PCR products were cleaned using an UltraClean PCR clean-up kit (MoBio Laboratories, Carlsbad, CA, USA) and fluorescently quantified using the Quant-iT PicoGreen dsDNA assay kit (Invitrogen, Carlsbad, CA, USA). Purified PCR products were pooled in equimolar concentrations and sequenced on a 2 × 150 bp Illumina MiSeq platform. Sequencing was conducted at the Microbiome Core, Steele Children Research Center at the University of Arizona.

Phyllosphere microbial cells were collected from leaflet surfaces using methods detailed by Kembel et al. (2014). The leaves were agitated for 5 min in a 1:50 diluted wash solution (1 M Tris·HCl, 0.5 M Na ethylenediaminetetraacetic acid, and 1.2% cetyltrimethylammonium bromide) (Kadivar and Stapleton, 2003) and pelleted via centrifugation at 4,000 g for 20 min. The supernatant was removed, and DNA was isolated using the same procedures applied to soil samples with the Qiagen DNeasy PowerLyzer PowerSoil Kit (Qiagen, Hilden, Germany) according to the manufacturers’ instructions. The same PCR amplification procedures, primer pairs, and sequencing methods used for soil samples were applied to phyllosphere samples.

Raw reads were demultiplexed using idemp (https://github.com/yhwu/idemp) and processed using DADA2 (Callahan et al., 2016) to assemble reads into error-corrected amplicon sequence variants (ASVs, phylotypes hereafter). The DADA2 pipeline included quality filtering, error rate modeling, dereplication, phylotype inference, merging of paired-end reads, chimera removal, construction of phylotype count table, and taxonomy assignment. Before quality filtering, cutadapt (Martin, 2011) was used to remove primer sequences in ITS reads due to length variation in the ITS region. During quality filtering, the 16S reads, but not the ITS reads, were truncated to the same length due to the potential for the truncation length to exceed ITS variant lengths. The 16S and ITS rRNA phylotypes were assigned taxonomic identities using the Ribosomal Database Project classifier (Wang et al., 2007) with a confidence threshold of 0.5, trained on the SILVA nr version 132 database (Quast et al., 2013) for bacteria/archaea and the UNITE database (Abarenkov et al., 2010) for fungi. The decontam package in R (R Core Team, 2020) was used to remove phylotypes identified as potential contaminants (a total of 12 ASVs in the negative controls) with the “prevalence” method at a threshold of 0.5 (Davis et al., 2018). Those 16S rRNA phylotypes without a bacterial or archaeal domain assignment and assigned to chloroplast or mitochondria were removed. Those ITS rRNA phylotypes without a fungal domain assignment were removed.

The putative functions of bacterial/archaeal phylotypes were inferred using FAPROTAX (Louca et al., 2016), and we focused on those functional groups associated with carbon (C) or nitrogen (N) cycling. The putative fungal guilds were inferred using FUNGuild (Nguyen et al., 2016), and the fungal phylotypes with a “highly probable” or “probable” confidence ranking assigned to a single guild were retained. We focused on those fungal guilds that can form pathogenic, symbiotic, or saprotrophic relationships with plants.

Statistical analyses were implemented in R (R Core Team, 2020). For each sample, the average phylotype richness was calculated after rarefying the phylotype count tables to the lowest number of reads per sample (6,160 and 11,675 for bacteria/archaea and fungi, respectively) for 100 iterations. Both bacterial/archaeal and fungal richness were normally distributed according to the Shapiro–Wilk test (bacteria/archaea: P = 0.23; fungi: P = 0.47). Bray–Curtis dissimilarity was measured after normalizing the phylotype count tables using cumulative-sum scaling (Paulson et al., 2013).

To compare microbial richness across sample types (i.e., soil and phyllosphere), land-use types, and locations, we used linear mixed-effects models which include sample types, land-use types, and locations (locations nested within land-use nested within sample type) as fixed effects and tree identity as a random effect. To compare microbial community composition across sample types, land-use types, and locations, a nested permutational multivariate analysis of variance was conducted (locations nested within land-use nested within sample type; 9,999 permutations). Homogeneity of multivariate dispersions was checked using the betadisper function in the vegan package. Post hoc contrasts were used to compare microbial richness and community composition in the soil versus phyllosphere, grazing versus natural, urban versus natural, as well as locations within the same land-use type. To compare the proportions of genera and inferred functional groups in grazing versus natural and urban versus natural areas, Wilcoxon rank-sum tests were performed. P values for multiple testing were corrected using the Benjamini–Hochberg method (Benjamini and Hochberg, 1995).

To compare the soil chemical properties across land-use types and locations, we fitted each of the soil chemical properties as a function of land-use types and locations (locations nested within land-use). Post hoc contrasts were used to compare soil chemical properties in grazing versus natural, urban versus natural, as well as locations within the same land-use type. To explore the environmental determinants of soil microbial richness and community composition in grazing versus natural, urban versus natural, as well as locations within the same land-use type, we fitted soil microbial richness (multiple linear regression) and community dissimilarity (multiple regression on distance matrices) as a function of soil chemical properties.

We used 2 methods to estimate the proportional contribution of soil to phyllosphere microorganisms. First, source tracking was executed using SourceTracker 2 (Knights et al., 2011), with phyllosphere samples being designated as “sink” and soil samples as “source.” Second, the proportion of phylotypes shared between the soil and phyllosphere (i.e., number of shared phylotypes divided by total number of phylotypes found in the soil and phyllosphere) was calculated. Both analyses were conducted for each mesquite tree separately. To compare these 2 proportions in grazing versus natural areas and urban versus natural areas, Wilcoxon rank-sum tests were performed.

Land-use type explained significant variations in all the soil chemical properties we examined (Table S1). Compared to natural areas, grazing areas had higher soil C and N contents, but these patterns were mainly driven by soil samples in the light grazing location (i.e., SR2; Figure S2A and B). Urban areas had higher soil C:N ratio (Figure S2C) and pH (Figure S2D) than natural areas. Location explained significant variations in soil C and N contents but not in C:N ratio, pH, and EC (Table S1). Within the natural and urban areas, all the soil chemical properties we examined were not significantly different among locations (Figure S3). However, within the grazing areas, soil C and N contents were significantly higher in SR2 than in SR1 (Figure S3).

After rarefaction, there were on average 633 bacterial/archaeal phylotypes (range = 193–851) and 209 fungal phylotypes (range = 68–313) detected per soil sample (Figure 1A and B). At the genus level, soil bacterial/archaeal communities were dominated by Pseudonocardia, Conexibacter, Streptomyces, Sphingomonas, and Solirubrobacter (Figure S4A). Soil fungal communities were dominated by Mortierella, Chaetomium, Aspergillus, Mycocentrospora, and Geastrum (Figure S4B). Compared to natural soils, urban soils had higher proportions of Nitrososphaera (archaeal genus) and Nocardiopsis (bacterial genus) but lower proportions of Acidisoma (bacterial genus) and Acidiphilium (bacterial genus) (Table S2). Compared to natural soils, grazing soils had higher proportions of Actinoallomurus (bacterial genus) and Cladophialophora (fungal genus) but lower proportions of Pseudomonas (bacterial genus) and Geosmithia (fungal genus) (Table S3).

On average, phyllosphere samples had 393 (range = 65–683) and 285 (range = 105–485) phylotypes for bacteria/archaea and fungi, respectively (Figure 1A and B). Phyllosphere bacterial/archaeal communities were predominately composed of taxa within Geodermatophilus, Hymenobacter, Microvirga, Sphingomonas, and Pseudonocardia (Figure S4C). The majority of phyllosphere fungi were members of Alternaria, Phaeococcomyces, Verrucoconiothyrium, Celosporium, and Aureobasidium (Figure S4D). Compared to natural areas, phyllosphere in urban areas had higher proportions of Microvirga (bacterial genus) and Alternaria (fungal genus) but lower proportions of Rickettsia (bacterial genus) and Leptospora (fungal genus) (Table S2). Compared to natural areas, phyllosphere in grazing areas had higher proportions of Pectobacterium (bacterial genus) and Mycosphaerella (fungal genus) but lower proportions of Labrys (bacterial genus) and Erythrobasidium (fungal genus) (Table S3).

Sample type explained significant variations in both microbial richness and community composition (Table 1). Phyllosphere samples had lower bacterial/archaeal richness but had higher fungal richness than soil (Figure 1A and B; Figure S5). In addition, soil and phyllosphere exhibited significantly different bacterial/archaeal and fungal community composition (Figure 1C and D). The dispersions of both bacterial/archaeal and fungal communities were significantly higher in the soil than in the phyllosphere, but they were not significantly different among locations and land-use types (Figure S6).

Land-use type explained significant variations in the richness of bacteria/archaea and fungi (Table 1). In the soil, the richness of bacteria/archaea and fungi was not significantly different between grazing and natural areas (Figure 2A and B; Figure S7), and their variations in grazing versus natural areas were not predictable by the soil chemical properties (Table S4). In the phyllosphere, the richness of fungi was higher in grazing than in natural areas, but the richness of bacteria/archaea was not significantly different between grazing and natural areas (Figure 2A and B; Figure S7).

In the soil, the richness of fungi was lower in urban than in natural areas (Figure 2A and B; Figure S7), and this pattern was mainly driven by the negative impacts of pH and EC on soil fungal richness in urban versus natural areas (Table S4). However, there was no significant difference in the richness of soil bacteria/archaea between urban and natural areas (Figure 2A and B; Figure S7), and none of the soil chemical properties was a significant predictor of the richness of soil bacteria/archaea in urban versus natural areas (Table S4). In the phyllosphere, the richness of both bacteria/archaea and fungi was higher in urban than in natural areas (Figure 2A and B; Figure S7).

Location explained a significant variation in the richness of fungi but not in bacteria/archaea (Table 1). Neither the soil bacterial/archaeal nor fungal richness was significantly different among locations within the same land-use type (Figure S8), although some of their variations can be explained by changes in soil properties (Table S5). Within the natural areas, the richness of phyllosphere bacteria/archaea and fungi was significantly higher in SC than in CSP, but neither of them was significantly different among locations in urban or grazing areas (Figure S8).

Land-use type explained significant variations in mesquite-associated soil and phyllosphere microbial community composition (Table 1). Specifically, the community composition of both soil- and phyllosphere-associated bacteria/archaea and fungi was significantly different in grazing versus natural and urban versus natural areas (Figure 2C and D). For grazing versus natural soils, the difference in soil bacterial/archaeal community composition was driven by change in pH, and the difference in soil fungal community composition was driven by changes in pH, EC, and N (Table S4). For urban versus natural soils, the differences in both soil bacterial/archaeal and fungal community composition were driven by changes in soil pH, EC, and C:N ratio (Table S4).

Location also explained significant variations in the community composition of bacteria/archaea and fungi (Table 1). Specifically, microbial community composition was significantly different among locations within the same land-use type, except for the soil bacterial/archaeal communities in CSP versus SC and RP versus UA (Table S6). Some of these patterns were driven by variations in soil chemical properties (Table S5). For example, the differences in both soil bacterial/archaeal and fungal communities in SR1 versus SR2 were predicted by changes in soil EC and N (Table S5).

We then explored if the proportions of functional groups vary across land-use types. We focused on functional groups of bacteria/archaea regulating C or N cycling as well as functional groups of fungi that can form pathogenic, symbiotic, or saprotrophic relationships with plants (Figure S9).

In the soil, grazing areas had a higher proportion of N-fixing bacteria (mainly Bradyrhizobium; Figure 3A) than natural areas. In the phyllosphere, grazing areas had a higher proportion of fungal plant pathogens (mainly Mycocentrospora and Aplosporella; Figure 3B) than natural areas.

In the soil, compared to natural areas, urban areas were associated with a higher proportion of fungal plant pathogens (mainly Mycocentrospora, Gibberella, and Thanatephorus; Figure 3A). However, urban areas exhibited a lower proportion of cellulolytic bacteria (mainly Isoptericola, Agromyces, and Acidothermus; Figure 3A).

In the phyllosphere, urban areas had higher proportions of fungal plant pathogens (mainly Mycocentrospora and Mycosphaerella), nitrifying bacteria (mainly Nitrososphaeraceae, Nitrosomonadaceae, and Nitrospiraceae), and denitrifying bacteria (mainly Paracoccus and Rhodoplanes; Figure 3B) than natural areas.

The proportion of phyllosphere microorganisms arising from soil (as estimated by SourceTracker 2) and the proportion of phylotypes shared between soil and phyllosphere were higher in urban than in natural areas, but they were not significantly different between grazing and natural areas for either bacteria/archaea or fungi (Figure 4).

Dryland ecosystems support approximately 78% of the global grazing area (Asner et al., 2004) and are home to over 38% of the human population (Maestre et al., 2012). These numbers are likely to increase due to the growing interest in pasture grazing and rapid population growth. Conversion of natural areas to grazing and urban lands can alter the diversity, community composition, and the functional capability of microorganisms (Hall et al., 2009; Byrnes et al., 2018; Lüneberg et al., 2018; Chen et al., 2021), which are key determinants of ecosystem processes and services in drylands (Maestre et al., 2015). Unlike previous studies (Laforest-Lapointe et al., 2017a; Schmidt et al., 2017), which only examine the changes in soil or phyllosphere microbiomes across land-use types, we simultaneously sampled both the soil and phyllosphere microbiomes of mesquite trees and compared them in grazing versus natural and urban versus natural areas. Overall, we showed that (i) mesquite-associated microbiomes were significantly different across land-use types, and these patterns were different between soil and phyllosphere and between bacteria/archaea and fungi; (ii) the proportions of some microbial functional groups varied significantly across land-use types; and (iii) the contribution of soil to phyllosphere microorganisms was higher in urban than in natural areas.

Soils harbor an immense number and diversity of microorganisms (Fierer, 2017). Although soil microorganisms can be aerosolized (Kellogg and Griffin, 2006; Schiro et al., 2022), only a small subset can colonize the phyllosphere (Massoni et al., 2021) because nutrient availability in the phyllosphere is limited as major carbon sources are simple sugars leached from the interior of the plant and leaf surfaces are subject to high ultraviolet radiation stress (Lindow and Brandl, 2003). Therefore, the community assembly of microorganisms in the phyllosphere is more likely to be governed by environmental selection, leading to the more homogenous microbial communities in the phyllosphere than in the soil. Moreover, it is not surprising that we observed a lower bacterial/archaeal richness in the phyllosphere than in the soil. These results are consistent with a previous study showing that the leaf surfaces of 2 perennial biofuel crops, switchgrass and miscanthus, had lower bacterial/archaeal richness than their associated soils (Grady et al., 2019). However, we found that fungal richness was higher in the phyllosphere than in the soil, which has rarely been reported in previous studies (Jia et al., 2020; Sun et al., 2021). A possible reason for the opposite patterns between bacteria/archaea and fungi is that soil-dwelling fungi are more likely to colonize the phyllosphere because they are more stress-tolerant than bacteria/archaea (Yuste et al., 2011; Allison et al., 2013). However, this possibility was excluded by our source tracker results, in which the proportional contribution of soil to phyllosphere microbiomes was not larger for fungi than for bacteria/archaea (Figure 4). We posit that other environmental sources such as atmosphere and insects might contribute to the higher fungal richness in the phyllosphere. Further studies conducted at larger scales and across more plant species are required to examine the generalizability of these results.

In the soil, our results showed that the richness of both bacteria/archaea and fungi was not significantly different between grazing and natural areas. These results are surprising because grazing has been found to alter soil microbial communities directly by modifying soil properties via animal trampling and manure deposition or indirectly by regulating plant productivity and community composition (Yang et al., 2019; Wang et al., 2021). In the arid Southwestern United States, cattle grazing can reduce the natural vegetative cover and soil organic matter, resulting in soil runoff and erosion (Byrnes et al., 2018; Condon et al., 2020). Therefore, we would expect that such changes in vegetation and soil properties would significantly alter the number of soil bacterial/archaeal and fungal phylotypes. Although our results showed that grazing soils had higher C and N contents than natural soils, there was no significant difference in soil microbial richness between grazing and natural areas, and none of the soil chemical properties was a significant predictor of soil microbial richness in grazing versus natural areas. These results suggest that microbial richness is a poor indicator of microbial response to grazing in arid ecosystems. However, previous studies have reported mixed results, with grazing being found to increase (Wang et al., 2022), decrease (Qu et al., 2016), or have no impact on soil microbial richness (Farrell et al., 2020).

In the soil, our results showed that the proportion of N-fixing bacteria was higher in grazing than in natural areas. Mesquite is a legume species and can form root nodules that are hosted by N-fixing bacteria, which can benefit the surrounding soil and neighboring plants (Fredrickson et al., 2006). It has been shown that grazing can alter soil N dynamics, specifically available N was lower in grazing areas than in natural areas due to reduced biomass of aboveground vegetation (Xie et al., 2021). This is particularly relevant in arid ecosystems because soil available N is a limiting factor (Ramond et al., 2022). Therefore, the higher proportion of N-fixing bacteria in grazing soils suggests that in rangelands mesquite canopies might be a promising avenue for restoration as they can serve as “nurse plants” that enhance beneficial relationships with neighboring plants (Padilla and Pugnaire, 2006; Gornish et al., 2021).

We also observed significant differences in soil microbial community composition between the heavy (i.e., SR1) and light grazing areas (i.e., SR2). Soil N content was higher in SR2 than in SR1, and it was a significant predictor of the differences in both soil bacterial/archaeal and fungal community composition in SR1 versus SR2. These results suggest that increasing grazing pressures can decrease soil N content and alter the soil microbial community composition. Moreover, an important implication of these results is that soil microbial communities are vulnerable to not only land-use change but also land-use intensity.

In the phyllosphere, the proportion of fungal plant pathogens was higher in grazing than in natural areas. Mesquite can be used as a feed resource for a wide range of livestock such as cattle, sheep, and goats (Sawal et al., 2004). Livestock might act as fungal vectors from soil and other plant species to the mesquite phyllosphere while foraging. Likewise, livestock foraging disturbances might indirectly contribute to the increased proportion of fungal plant pathogens in the phyllosphere. Specifically, fungal plant pathogens accounted for approximately 50% of the sequences in one of the phyllosphere sample collected from the grazing areas (Figure 3B), suggesting that there might be intensive livestock foraging on that mesquite tree. These results suggest that grazing might influence the growth and health of mesquite by increasing pathogen loads.

Organisms may respond differently to urbanization due to their physiological differences (Mcdonald et al., 2008). On the one hand, urban avoiders (i.e., those species reach their highest densities at the most natural sites; Blair, 1996) can be excluded by increased cover of impervious surfaces combined with other urban stressors like compaction and pollution, resulting in a decrease in species richness (Gill et al., 2020). On the other hand, richness might increase with increasing urban intensity because urban exploiters (i.e., those species reach their highest densities in highly modified habitats; Blair, 1996) can perceive urban environments as ecological opportunities and are able to exploit resources and attain large population densities (Liu et al., 2022). Therefore, the response of microbial richness to urbanization depends on the balance between urban avoiders and exploiters (Dornelas et al., 2019).

Our results showed that urban soils had higher proportions of Nitrososphaera and Nocardiopsis, which can be classified as urban exploiters. In contrast, urban soils had lower proportions of Acidiphilium and Acidisoma, which can be classified as urban avoiders. These results can be attributed to the higher soil pH in urban soils. It has been shown that Nitrososphaera (Bates et al., 2011) and Nocardiopsis (Hozzein et al., 2004) tend to be prevalent in alkaline soils, and Acidiphilium and Acidisoma are acidophiles (Kuang et al., 2013). Moreover, we found that Rickettsia, which is an insect-associated bacterial genus (Engel and Moran, 2013), was an urban avoider. This genus was less abundant in the phyllosphere of urban areas because insect diversity is expected to be lower in urban than in natural areas.

In this study, we sampled soils underneath mesquite canopies and found that fungal richness was lower in urban than in natural areas, but bacterial/archaeal richness was not significantly different between urban and natural areas. Similar patterns were observed in a previous study, where we sampled bare soils and showed that soil fungal richness decreased but soil bacterial/archaeal richness did not change across an urbanization gradient in the arid Southwestern United States (Chen et al., 2021). These results suggest that in drylands plant-derived nutrient inputs cannot alleviate the negative impacts of urbanization on soil fungal richness and that soil bacterial/archaeal richness is insensitive to urbanization whether they are exposed or protected by mesquite canopies. In drylands, plant cover tends to be low and is a limiting factor for ecosystem productivity (Maestre et al., 2012). It is possible that there may exist more urban avoiders than exploiters for soil fungi because they require considerable plant cover, which might be limited in urban environments. In contrast, soil bacteria/archaea might be more tolerant to low plant cover and other urban stressors, and thus, the loss of urban avoiders can be compensated by the gain of urban exploiters. Moreover, horizontal gene transfer can increase the resistance of bacteria/archaea to urban stressors, but it might be less frequent in fungi (Van Etten and Bhattacharya, 2020).

For mesquite-associated phyllosphere microbiome, our results showed that the richness of both bacteria/archaea and fungi was higher in urban than in natural areas, which is consistent with a previous study showing that phyllosphere microbial richness increased along an urban intensity gradient (Laforest-Lapointe et al., 2017a). In urban environments, there might be increased particulate matter caused by heavy traffic pollution (Imperato et al., 2019) or higher concentrations of micronutrients and macronutrients caused by atmospheric pollutants (Perreault and Laforest-Lapointe, 2022). These factors can promote the colonization and establishment of urban exploiters in the phyllosphere, resulting in a net increase in microbial richness.

In both the soil and phyllosphere, our results showed that the proportion of fungal plant pathogens was higher in urban than in natural areas, suggesting that fungal diseases might be a bigger threat to mesquite health in urban environments. Future studies aiming at exploring how the increased pathogen loads affect the growth and health of mesquite in urban landscapes can provide valuable insights into the management of urban mesquite trees, especially as mesquite can be used as food crop, fuel wood, and commercial hardwood products, as well as for restoration in drylands (Nabhan et al., 2020; Gornish et al., 2021).

In the phyllosphere, we found that both the proportions of nitrifying and denitrifying bacteria were higher in urban than in natural areas. The leaf surface of trees constantly receives N via atmospheric deposition (Guerrieri et al., 2015), which might be higher in urban environments because of the growing rates of fossil fuel combustion and use of N-based fertilizers, ultimately leading to an increase in the proportions of taxa associated with N cycling.

The phyllosphere microbiome is the combined product of various environmental sources and is generally subject to high variability due to the inherently open nature of the environment (Bao et al., 2020), with soil being often found to be the primary source (Vacher et al., 2016). Interestingly, our results showed that the contribution of soil to phyllosphere microorganisms was higher in urban than in natural areas. In urban locations, with relatively loose soils, construction activity, as well as vehicle and pedestrian traffic might facilitate the aerosolization and dust dispersal of soil microorganisms (Duniway et al., 2019), and the aridity in drylands can promote these soil microorganisms to reach the phyllosphere via dry deposition.

We found that the richness, composition, and functional potential of mesquite-associated microbiomes changed significantly across land-use types although these changes were different between soil and phyllosphere, as well as between bacteria/archaea and fungi. It is worth noting that, however, our sampling was conducted at a local scale and the functional potential of microbial taxa was inferred by their taxonomy. Future studies pairing large-scale sampling with quantitative molecular approaches are required to test the generality of our results. Our results provide an initial foundation to understand how different land uses shape the diversity and functioning of terrestrial ecosystems, and in particular, the impacts of urban and grazing areas on the growth and health of mesquite trees with potential applications for urban conservation gardening (Segar et al., 2022), food production in drylands (Nabhan et al., 2020), and rangeland management and restoration (Gornish et al., 2021).

The raw sequencing data have been deposited in the NCBI Sequence Read Archive under BioProject accession code PRJNA839634. Soil chemical data have been deposited in Figshare: https://doi.org/10.6084/m9.figshare.23615547.

The supplemental files for this article can be found as follows:

Tables S1 to S6. Figures S1 to S9.PDF

The authors thank Gabriele Schiro and Joseph C. Blankinship for help with soil chemical analyses. An allocation of computer resources from the UA Research Computing High Performance Computing (HPC) at the University of Arizona is gratefully acknowledged.

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

The authors declare no conflict of interest.

Conceived the study: AB.

Conducted sampling and laboratory analyses: SC, YC, AB.

Performed data analyses: SC, YC.

Contributed to the writing of the manuscript and assisted with revisions: All authors.

How to cite this article: Cleavenger, S, Chen, Y, Barberán, A. 2023. Mesquite-associated soil and phyllosphere microbial communities differ across land-use types in drylands. Elementa: Science of the Anthropocene 11(1). DOI: https://doi.org/10.1525/elementa.2023.00026

Domain Editor-in-Chief: Steven Allison, University of California Irvine, Irvine, CA, USA

Associate Editor: Stephanie Kivlin, Department of Ecology and Evolutionary Biology, University of Tennessee Knoxville, Knoxville, TN, USA

Knowledge Domain: Ecology and Earth Systems